GTM-MQNC2Z4
Skip to main content
  1. Home >
  2. About Fujitsu >
  3. Environment >
  4. Our Customers and Society >
  5. Co-creation2: A Traffic-Monitoring System That Uses AI-Driven Image-RecognitionTechnologies

Co-creation2: A Traffic-Monitoring System That Uses AI-Driven Image-RecognitionTechnologies

Cutting CO2 emissions by enabling accurate detection of congestion and accidents

Traffic congestion is becoming an increasingly serious social problem in many countries, especially emerging nations. Not only do crowded streets hamper economic development, but congestion also endangers the environment. To help mitigate these traffic issues, Fujitsu developed a traffic-monitoring system that uses AI-driven image-recognition technologies to accurately detect road congestion and accidents in real time.

Cutting CO2 emissions by enabling accurate detection of congestion and accidents

In emerging nations, burgeoning populations and fast-moving urbanization are creating the social issue of traffic congestion. Congestion does more than simply impede economic growth; it also hurts the natural environment by contributing to air pollution, and climate change. To help mitigate these traffic issues, Fujitsu developed a traffic-monitoring system that uses image-recognition technologies̶which draw on Zinrai, Fujitsu's unique AI technology̶to accurately detect road congestion and accidents in real time.

The issue with conventional image-recognition technologies was that they were affected by the environmental, including headlights, sunlight, and shadows, which made it hard to analyze video from monitoring cameras. It was also difficult to efficiently and accurately recognize such varied and complex incidents as traffic accidents and violations.

Fujitsu's technology boosts recognition accuracy through AI technology, which learns a large, pre-assembled library of images. Leveraging that asset, Fujitsu's new technology can identify vehicle positions̶even at night or under foggy conditions. The technology also observes changes in vehicle movement and speed. The technology quantifies the corresponding anomaly to make the process of detecting accidents and violations more efficient. In 2015, Fujitsu conducted collaborative field trials with the Tsingha University Suzhou Automobile Research Institute to test the technology in China. The joint effort achieved high levels of recognition accuracy at 90% to 95%.

With the new technology, users can gather congestion-related information that not only provides insight into the strategies for placing and controlling traffic lights but also helps identify ways of curbing the accidents and violations, leading to improvements in many areas, including safer urban transportation and fewer CO2 emissions. Organizations in Japan, China, and Europe are planning to implement the system and start using the service, which Fujitsu will launch services for in FY 2017.

Detecting a wide variety of conditions in real time

Enabling efficient, centralized management of conditions and incidents on a large geographical scale